In a cluster computing environment multiple computer nodes typically share data storage devices. In order to prevent conflicts that may arise when multiple nodes in a cluster attempt to access the same data storage device at the same time, techniques have been developed to provide access to data storage devices in a controlled manner, such as where a node is given exclusive ownership over part or all of a data storage device for a period of time. For example, in a SCSI-based Storage Area Network (SAN), a node may issue a “SCSI reserve” command to obtain exclusive ownership of a SCSI device. Once a SCSI device is reserved in this manner, it refuses all commands from all other nodes until exclusive ownership of the SCSI device is revoked, such as after an “SCSI reset” command is received by the SCSI device.
Back-up copies of data stored on data storage devices, such as for disaster-recovery purposes, are often made without taking the data storage devices off-line, thus ensuring that primary data providers/consumers, such as computer users and other processes and applications, have continuous access to data. Such back-ups are often made by dedicated back-up servers that access data storage devices by making data requests of the nodes in a cluster that are connected to the data storage devices. Unfortunately, this results in additional processing overhead for such participating nodes. Alternatively, dedicated back-up servers access data storage devices via data communications channels that are separate from those that are normally used to provide data to primary data providers/consumers. This is done to minimize the impact that the data back-up service has on the performance and responsiveness of the overall system. Unfortunately, in a cluster computing environment where the nodes themselves control access to data storage devices, a dedicated back-up server may be denied access to a data storage device, thus hindering the data back-up service.